{ "bcoid": "GLY_000022", "title": "Mouse Protein Information (NCBI RefSeq)", "description": "The Mouse NCBI RefSeq Protein Information dataset contains mouse [taxid:10090] protein information such ...", "filename": "mouse_protein_info_refseq.csv", "filetype": "csv", "categories": { "species": "Mus Musculus", "molecule": "Protein", "tags": "Protein Information", "file_type": "csv", "status": "Reviewed", "scope": "internal" }, "minitable": { "content": [ [ "Q61527-1...", "XP_006495755..." ], [ "P25425-1...", "XP_006496767..." ] ], "headers": [ "uniprotkb_ca...", "p_refseq_ac_..." ], "colwidth": [ "40%", "20%" ] }, "sampledata": { "type": "table", "data": [ [ { "type": "string", "label": "uniprotkb_canonical_ac" }, { "type": "string", "label": "p_refseq_ac_best_match" }, { "type": "string", "label": "refseq_protein_name" }, { "type": "string", "label": "refseq_protein_length" }, { "type": "string", "label": "refseq_protein_summary" } ], [ "Q61527-1", "XP_006495755.1", "receptor tyrosine-protein kinase erbB-4 isoform X1", "1308", "" ], [ "P25425-1", "XP_006496767.1", "POU domain, class 2, transcription factor 1 isoform X1", "770", "" ], [ "B2RX88-1", "XP_006495545.1", "centrosome and spindle pole associated protein 1 isoform X4", "1205", "" ], [ "B9EKE5-1", "XP_006496047.1", "ciliogenesis-associated TTC17-interacting protein isoform X3", "520", "" ], [ "Q6A065-1", "XP_006497000.1", "centrosomal protein of 170 kDa isoform X6", "1588", "" ], [ "D3Z5G0-1", "XP_011237154.1", "uncharacterized protein LOC623121 isoform X3", "586", "" ], [ "Q91WJ7-1", "XP_006496278.1", "SPATS2-like protein isoform X4", "558", "" ], [ "Q5DTI6-1", "XP_006496304.1", "KAT8 regulatory NSL complex subunit 1-like protein isoform X3", "991", "" ], [ "A0A087WPV9-1", "XP_011236891.1", "uncharacterized protein LOC69397 isoform X3", "146", "" ], [ "D3Z7A5-1", "XP_006496611.1", "dynein assembly factor with WDR repeat domains 1 isoform X2", "415", "" ], [ "A0A286YDK0-1", "XP_999090.1", "uncharacterized protein 3110040M04Rik", "300", "" ], [ "A0A140LIP2-1", "XP_017169236.1", "epithelial cell-transforming sequence 2 oncogene-like isoform X1", "953", "" ], [ "Q60636-1", "XP_006512568.1", "PR domain zinc finger protein 1 isoform X4", "856", "" ], [ "P17710-1", "XP_006513305.1", "hexokinase-1 isoform X1", "974", "" ], [ "Q6A078-1", "XP_006513591.1", "centrosomal protein of 290 kDa isoform X3", "2472", "" ], [ "Q3T9Z9-1", "XP_006512931.1", "zinc finger with UFM1-specific peptidase domain protein isoform X1", "577", "" ], [ "Q8K400-1", "XP_006512989.1", "syntaxin-binding protein 5 isoform X1", "1152", "" ], [ "A0A1W2P7G7-1", "XP_011241916.1", "zinc finger protein 431-like", "749", "" ], [ "A0A1W2P6H7-1", "XP_006514422.2", "zinc finger protein 431-like isoform X3", "600", "" ], [ "J3QMW1-1", "XP_001477836.1", "keratin-associated protein 10-8 isoform X1", "236", "" ], [ "L7N2C3-1", "XP_006514415.1", "keratin-associated protein 10-8-like", "230", "" ], [ "A0A1W2P6U8-1", "XP_006514423.1", "uncharacterized protein LOC102635990 isoform X1", "4986", "" ], [ "Q5SNZ0-1", "XP_006514518.1", "girdin isoform X6", "1873", "" ], [ "Q5SUF7-1", "XP_006532133.1", "voltage-dependent T-type calcium channel subunit alpha-1G isoform X4", "2381", "" ], [ "P70351-1", "XP_006532241.1", "histone-lysine N-methyltransferase EZH1 isoform X1", "747", "" ], [ "Q5NCJ1-1", "XP_006532624.1", "rap guanine nucleotide exchange factor 6 isoform X2", "1609", "" ], [ "A2A863-1", "XP_006532635.1", "integrin beta-4 isoform X4", "1818", "" ], [ "Q5SSM3-1", "XP_006532934.1", "rho GTPase-activating protein 44 isoform X1", "814", "" ], [ "A9XX86-1", "XP_006532975.1", "macrophage galactose N-acetyl-galactosamine specific lectin 2 isoform X1", "381", "" ], [ "Q5SV77-1", "XP_011247235.1", "gametogenetin-binding protein 2 isoform X4", "696", "" ], [ "Q5SVQ0-1", "XP_006533076.1", "histone acetyltransferase KAT7 isoform X1", "613", "" ], [ "A0A140LIY9-1", "XP_011247256.1", "glutamine-rich protein 2 isoform X1", "2337", "" ], [ "B1AR10-1", "XP_006533395.1", "protein AF-17 isoform X1", "1109", "" ], [ "Q5F259-1", "XP_006533458.1", "ankyrin repeat domain-containing protein 13B isoform X11", "626", "" ], [ "Q9JMH9-3", "XP_006533667.1", "unconventional myosin-XVIIIa isoform X12", "2050", "" ], [ "J3QK56-1", "XP_006534029.1", "uncharacterized protein C17orf107 homolog isoform X1", "214", "" ], [ "B2RUP2-1", "XP_006534251.1", "protein unc-13 homolog D isoform X1", "1085", "" ], [ "Q8K4R4-1", "XP_006534325.1", "cytoplasmic phosphatidylinositol transfer protein 1 isoform X1", "332", "" ], [ "A2A9T0-1", "XP_006534398.1", "TANK-binding kinase 1-binding protein 1 isoform X1", "611", "" ], [ "Q5SW19-1", "XP_006534416.1", "clustered mitochondria protein homolog isoform X2", "1315", "" ], [ "Q8K296-1", "XP_006514910.1", "myotubularin-related protein 3 isoform X2", "1196", "" ], [ "Q5NC63-1", "XP_017170354.1", "zinc finger protein 717-like", "627", "" ], [ "A0A1B0GT45-1", "XP_011247669.1", "protein PRRC2C", "1251", "" ], [ "A0A286YDN8-1", "XP_017170386.1", "ankyrin repeat domain-containing protein 40-like", "145", "" ], [ "P70365-1", "XP_006515068.1", "nuclear receptor coactivator 1 isoform X1", "1447", "" ], [ "Q02384-1", "XP_006515698.1", "son of sevenless homolog 2 isoform X1", "1333", "" ], [ "Q6NS52-1", "XP_017170508.1", "diacylglycerol kinase beta isoform X2", "802", "" ], [ "Q8R0W1-1", "XP_011242380.1", "CLOCK-interacting pacemaker isoform X2", "432", "" ], [ "A0A1Y7VMI0-1", "XP_006515276.1", "leucine-rich repeat-containing protein 72 isoform X1", "288", "" ], [ "A0A1Y7VJE9-1", "XP_006516499.1", "eukaryotic translation initiation factor 1A-like", "144", "" ], [ "Q3U2K0-1", "XP_006517257.1", "protein FAM193B isoform X1", "892", "" ], [ "F6R3P9-1", "XP_006516702.1", "prolactin family 3, subfamily d, member 1 isoform X1", "224", "" ], [ "Q5SZA1-1", "XP_006516711.1", "sodium-dependent phosphate transport protein 3 isoform X2", "478", "" ], [ "Q9ESN5-1", "XP_017171041.1", "centromere protein K isoform X2", "271", "" ], [ "Q3UH06-1", "XP_006516822.1", "ras-responsive element-binding protein 1 isoform X3", "1700", "" ], [ "Q3UHX0-1", "XP_006517019.2", "nucleolar protein 8 isoform X1", "1147", "" ], [ "Q3U827-1", "XP_006517822.1", "E3 ubiquitin-protein ligase RNF180 isoform X1", "592", "" ], [ "E9PV86-1", "XP_017171131.1", "multiple C2 and transmembrane domain-containing protein 1 isoform X4", "951", "" ], [ "J3KMS6-1", "XP_011243019.1", "uncharacterized protein Gm21818", "170", "" ], [ "A0A286YCZ6-1", "XP_011242665.1", "uncharacterized protein C6orf229 homolog", "238", "" ], [ "F6V5V4-1", "XP_619089.4", "serpin B9", "377", "" ], [ "Q8BRT5-1", "XP_001473814.1", "uncharacterized protein A830005F24Rik", "167", "" ], [ "A0A1Y7VN40-1", "XP_006517059.1", "spermatogenesis-associated protein 31A6-like", "1110", "" ], [ "A0A1Y7VIZ1-1", "XP_017171165.1", "zinc finger protein 431-like isoform X3", "466", "" ], [ "J3QNN2-1", "XP_006517862.1", "low-density lipoprotein receptor class A domain-containing protein 1-like", "249", "" ], [ "E9Q5P1-1", "XP_017171209.1", "predicted gene 3164 isoform X1", "227", "" ], [ "A0A286YCV9-1", "XP_006518682.1", "kinesin-like protein KIF13B isoform X1", "1847", "" ], [ "Q8BYJ6-1", "XP_006518822.1", "TBC1 domain family member 4 isoform X1", "1307", "" ], [ "Q9QYJ7-1", "XP_006519139.1", "dual specificity protein phosphatase 13 isoform X2", "198", "" ], [ "Q8BRB7-1", "XP_006519331.1", "histone acetyltransferase KAT6B isoform X3", "1872", "" ], [ "A0A2I3BR81-1", "XP_006519570.1", "5'-nucleotidase domain-containing protein 2 isoform X1", "553", "" ], [ "Q8K2Y0-1", "XP_011243508.1", "RING finger protein 219 isoform X1", "214", "" ], [ "L7N2A3-1", "XP_001472902.3", "uncharacterized protein Gm2244", "218", "" ], [ "A0A2I3BPJ2-1", "XP_017171734.1", "anthrax toxin receptor-like isoform X1", "573", "" ], [ "F7BTS7-1", "XP_017171740.1", "uncharacterized protein LOC108168148 isoform X2", "198", "" ], [ "E9PZI8-1", "XP_017171750.1", "uncharacterized protein Gm7233 isoform X2", "198", "" ], [ "K7N744-1", "XP_011243571.1", "uncharacterized protein LOC101056159 isoform X2", "198", "" ], [ "E9Q2M1-1", "XP_017171755.1", "uncharacterized protein LOC108168155 isoform X1", "201", "" ], [ "L7N2D2-1", "XP_001476726.3", "uncharacterized protein Gm3278", "218", "" ], [ "L7N2E2-1", "XP_011243108.1", "uncharacterized protein Gm8271", "218", "" ], [ "D3Z7A4-1", "XP_011243138.1", "uncharacterized protein LOC105245675", "205", "" ], [ "A0A1B0GRK0-1", "XP_017171840.1", "uncharacterized protein LOC108168208", "198", "" ], [ "Q07916-1", "XP_011243814.1", "POU domain, class 6, transcription factor 1 isoform X8", "301", "" ], [ "Q3U0M1-1", "XP_006521590.1", "trafficking protein particle complex subunit 9 isoform X1", "1148", "" ] ] }, "readme": "Name\n - Mouse Protein Information (NCBI RefSeq)\n\nVersion\n - 1.4.5\n\nCreated on\n - 2019-01-23T17:21:05.888176\n\nModified on\n - 2019-09-12T17:15:46.889793\n\nCreated by\n - Jeet Vora [jeetvora@gwu.edu], Rahi Navelkar [rsn13@gwu.edu], Robel Kahsay [rykahsay@gwu.edu], Xavier Holmes [xavierh@email.gwu.edu]\n\nAuthors\n - \n\nChecksum\n - \n\nUsability Domain\n - The Mouse NCBI RefSeq Protein Information dataset contains mouse [taxid:10090] protein information such as protein accession, name, length and protein function summmary from NCBI RefSeq database. If you use this dataset please provide proper attribution to NCBI RefSeq and GlyGen\n\nDescription Domain\n Keywords\n - protein information\n - protein length\n - protein summary\n - NCBI RefSeq\n\n Pipeline Steps\n - Step-1: The script parses the NCBI RefSeq gpff files to extract the accession, length, protein summary and then maps the NCBI RefSeq accesion to the UniProtKB Canonical accessions\n\nExecution Domain\n Scripts - https://github.com/GW-HIVE/glygen-backend-integration/blob/master/pipeline/integrator/make-protein-dataset.py\n\n\n Software Prerequisites\n Name: Python\n Version: 2.7.5\n URI: https://www.python.org/download/releases/2.7.5/\n\nI/O Domain\n Input Subdomain\n Name: refseq_protein_all_10090.gpff\n URI: https://data.glygen.org/ln2wwwdata/source/refseq_protein_all_10090.gpff\n Name: mouse_protein_xref_refseq.csv\n URI: https://data.glygen.org/ln2wwwdata/reviewed/mouse_protein_xref_refseq.csv\n Output Subdomain\n Name: mouse_protein_info_refseq.csv\n URI: http://data.glygen.org/ln2wwwdata/reviewed/mouse_protein_info_refseq.csv\n Name: mouse_protein_info_refseq.stat.csv\n URI: http://data.glygen.org/ln2wwwdata/reviewed/mouse_protein_info_refseq.stat.csv\n\nFields\n Name: \"refseq_protein_summary\"\n Description: \"Function of the protein in NCBI RefSeq database\"\n Unique Values: \"923\"\n Name: \"uniprotkb_canonical_ac\"\n Description: \"Accession assigned to the protein isoform chosen to be the canonical sequence to which all positional annotation refers in UniProtKB database\"\n Unique Values: \"14869\"\n Name: \"p_refseq_ac_best_match\"\n Description: \"Best match NCBI RefSeq protein accession to UniProtKB canonical accession\"\n Unique Values: \"14869\"\n Name: \"refseq_protein_name\"\n Description: \"Name of the protein in NCBI RefSeq database\"\n Unique Values: \"14780\"\n Name: \"refseq_protein_length\"\n Description: \"Length of the protein in NCBI RefSeq database\"\n Unique Values: \"1858\"", "downloadurl": "/ln2data/releases/data/v-1.4.5/reviewed/mouse_protein_info_refseq.csv" }